Relative accuracy of emergency CT in adults with non-traumatic abdominal pain

CT examination prior to emergency laparotomy has become ubiquitous in contemporary clinical practice, but the relative accuracy of CT in this context has not been widely reported. The aim of this study was to determine the accuracy and strength of agreement between the perceived pre-operative CT diagnosis and operative findings.

Data from patients undergoing pre-operative CT prior to emergency laparotomy from January 2013 to June 2014 in a large teaching hospital were analysed. The CT diagnosis was compared with operative findings using the χ(2) test and weighted kappa statistic (Kw). Results were further analysed related to the time of day the CT was reported, anatomical location and grade of the reporting radiologist.

361 patients [median age 67 years (18-98 years); 180 males] underwent CT prior to emergency laparotomy. CT reports were deemed accurate in 318 (88.1%) cases and inaccurate in 43 (11.9%) cases, which resulted in 5 negative laparotomies in this latter cohort (11.6%, χ(2) 37.50, df 1; p < 0.0001). Accuracy and strength of agreement varied with anatomical location of the pathology; upper gastrointestinal (UGI) 75.5%, Kw 0.673 (0.531-0.815; p < 0.001); small bowel 89.9%, Kw 0.781 (0.687-0.875, p < 0.001); lower gastrointestinal (LGI) 90.4%, Kw 0.821 (0.749-0.893; p < 0.001). CT examinations reported within normal working hours had higher strength of agreement [Kw 0.832 (0.768-0.896), p < 0.001] than CTs reported out of hours [Kw 0.789 (0.721-0.857), p < 0.001], but there was no significant difference in overall accuracy (89.9 vs 86.0%; χ(2) 1.306, df 1, p = 0.253). Reporter seniority was not associated with improved diagnostic accuracy (χ(2) 1.825, df 1; p = 0.177).

CT agreement with emergency operative pathology was good to excellent, but the strength of agreement varied in relation to anatomical location of pathology.

Overall accuracy was 88.1% with good to excellent agreement between pre-operative CT and emergency laparotomy findings in adult patients with non-traumatic abdominal pain in the acute setting. Diagnostic accuracy of CT reporting varies with anatomical location of pathology.